Neural mechanisms of phonetic categorization analyzed based on MMF elicited by acoustic and phonemic variations

Humans have a superior capability to categorize phonetic sounds against non-linguistic speaker-dependent acoustic variations. Neural mechanisms for such a stable phonetic categorization were analyzed based on the mismatch magnetic fields (MMFs) measurements elicited by acoustic and phonemic variations. The MMFs were analyzed comparing within-category (WC) versus across-category (AC) variations under multi-speaker (MS) or single-speaker (SS) conditions. With SS, the equivalent current dipole (ECDs) moments of MMF elicited by AC were significantly larger than those by WC in the left hemisphere (LH), and ECDs by WC were significantly larger than those by AC in the right hemisphere (RH). With MS, ECDs by AC were larger than those by WC in the LH. In the RH, however, ECDs by AC were the same as those by WC. These results suggest that the MMF of the LH reflects phonetic categorization as well as detection of acoustic variations, while that of the RH reflects mainly the latter.